io.github.nuance-dev/rival

Install

Config snippet generator goes here (5 client tabs)

README

<p align="center">
  <img src="https://www.rival.tips/favicon.svg" width="64" height="64" alt="RIVAL" />
</p>

<h1 align="center">rival-mcp</h1>

<p align="center">
  MCP server for querying AI model comparison data from <a href="https://www.rival.tips">rival.tips</a>
</p>

This server lets AI coding assistants — Claude Code, Cursor, Windsurf, and any MCP-compatible client — natively query model benchmarks, pricing, capabilities, and side-by-side comparisons without leaving your editor.

## Quick Start

```bash
npx rival-mcp
```

No API key required. All data is served from the public rival.tips API.

## Configuration

### Claude Code

Add to your `.claude/settings.json` (project-level) or `~/.claude/settings.json` (global):

```json
{
  "mcpServers": {
    "rival": {
      "command": "npx",
      "args": ["-y", "rival-mcp"]
    }
  }
}
```

### Claude Desktop

Add to `~/Library/Application Support/Claude/claude_desktop_config.json` (macOS) or `%APPDATA%\Claude\claude_desktop_config.json` (Windows):

```json
{
  "mcpServers": {
    "rival": {
      "command": "npx",
      "args": ["-y", "rival-mcp"]
    }
  }
}
```

### Cursor

Add to your Cursor MCP settings (`.cursor/mcp.json`):

```json
{
  "mcpServers": {
    "rival": {
      "command": "npx",
      "args": ["-y", "rival-mcp"]
    }
  }
}
```

### Windsurf

Add to your Windsurf MCP config:

```json
{
  "mcpServers": {
    "rival": {
      "command": "npx",
      "args": ["-y", "rival-mcp"]
    }
  }
}
```

## Available Tools

### `list-models`

List all AI models with optional filtering.

**Parameters:**
| Parameter | Type | Description |
|-----------|------|-------------|
| `provider` | string (optional) | Filter by provider: `OpenAI`, `Anthropic`, `Google`, `Meta`, `Mistral`, etc. |
| `category` | string (optional) | Filter by category: `flagship`, `reasoning`, `coding`, `small`, `free`, `image-gen` |
| `capability` | string (optional) | Filter by capability: `chat`, `code`, `vision`, `image-gen`, `function-calling` |
| `q` | string (optional) | Free-text search across name, ID, provider, and description |

**Example prompts:**
- "List all Anthropic models"
- "Show me free models"
- "What models support vision?"

### `get-model`

Get detailed information about a specific model — benchmarks, pricing, capabilities, unique features, and provider availability.

**Parameters:**
| Parameter | Type | Description |
|-----------|------|-------------|
| `id` | string (required) | Model ID, e.g. `gpt-4.1`, `claude-3.7-sonnet`, `gemini-2.5-pro` |

**Example prompts:**
- "Get details on claude-3.7-sonnet"
- "What are the benchmarks for gpt-4.1?"

### `compare-models`

Compare 2-3 models side by side — benchmarks, pricing, capabilities, and shared challenges.

**Parameters:**
| Parameter | Type | Description |
|-----------|------|-------------|
| `models` | string (required) | Comma-separated model IDs (2-3). Example: `gpt-4.1,claude-3.7-sonnet` |

**Example prompts:**
- "Compare GPT-4.1 vs Claude 3.7 Sonnet"
- "How does Gemini 2.5 Pro stack up against GPT-4.1 and Claude Sonnet?"

### `search-models`

Search for models by name, description, or capability when you don't know the exact model ID.

**Parameters:**
| Parameter | Type | Description |
|-----------|------|-------------|
| `query` | string (required) | Search query, e.g. `vision`, `cheap coding`, `fast reasoning` |

**Example prompts:**
- "Find models good at coding"
- "Search for cheap reasoning models"

## Development

```bash
# Install dependencies
npm install

# Run in development mode
npm run dev

# Build for production
npm run build

# Run the built server
npm start
```

## How It Works

This MCP server communicates over stdio (standard input/output) using the [Model Context Protocol](https://modelcontextprotocol.io). When an AI assistant needs model comparison data, it calls the appropriate tool, which fetches data from the rival.tips public API and returns structured JSON.

The server exposes no resources or prompts — only tools. All data is read-only and publicly available.

## Data Source

All model data comes from [rival.tips](https://www.rival.tips), an AI model comparison platform featuring:

- 60+ AI models with benchmarks, pricing, and capability data
- Side-by-side comparisons with shared challenge responses
- Community-driven AI duel voting and rankings
- Pre-generated showcase responses across coding, creative, and reasoning tasks

## License

MIT